Agent Judgment Simulator Site audit findings flow through a decision tree. Agents should flag by default — auto-fix only when the fix is mechanical, reversible, and high-confidence.
Run All Reset
Tap a finding to trace its decision path
Broken link to archived blog post broken-link Link returns 404 but archive.org has the content at a known URL. React version one minor behind stale-dependency 19.0.0 installed, 19.1.0 available. Breaking changes possible. Blog post references "Next.js 15" as latest content Content is stale but the wording might be intentional context. Missing og:image on /lab page meta No social preview image. Low SEO impact for an internal lab page. Typo in a blog post heading content "Authentiation" instead of "Authentication." Clear misspelling. npm audit: low-severity vuln in dev dependency security No user impact. Development-only tooling with known advisory. Lighthouse performance score dropped to 72 performance Performance regression detected. Multiple potential root causes. Copyright year says 2025 content Footer still shows last year. Straightforward date update. External SDK URL changed (301 redirect) broken-link Link works but is redirected. New URL might change again. Console error from third-party analytics script performance No user-visible impact. Script loads but logs a non-fatal error.
YES NO YES NO YES NO YES NO YES NO Mechanical fix? Reversible? Confidence > 95%? Affects users? Cosmetic only? Auto-fix Flag for Review Ignore Finding enters ✅ Auto-fix 0
No findings yet
📋 Flag for Review 0
No findings yet
✅ 0 📋 0 ⊘ 0